The 18-Month Window for Construction Workers

In this episode of Construction Conversations, I sat down with Dr. Sarah Buchner, CEO and Founder of Trunk Tools, to talk about what it takes to build a construction AI company from scratch, why most C-suite leaders still aren’t using AI themselves, and what the industry risks losing if it automates the wrong things.

Sarah’s path into construction tech wasn’t conventional. She grew up on a farm in Austria, started working as a carpenter at 12, became a superintendent, then pivoted to software after experiencing a fatality on a jobsite she was running. That trajectory, from blue collar roots to a PhD in construction data to leading a Series B AI company, informs everything about how she thinks about building for the field.

But here’s what matters right now. Sarah believes there’s an 18-month window where the people who move first will have an advantage that doesn’t come back. After that, AI adoption becomes normal. Your company mandates it. The edge disappears. This is what to do with that window.

Here’s what stood out.

1. The Field Is Still the North Star, Even When the Market Pulls You Elsewhere.

Sarah was blunt about the tension she lives with daily. Her instinct is to build for the field, for superintendents, for the people running projects. But enterprise customers keep pushing her team toward back-office automation and white-collar workflows. “I keep wanting to build for the field and we keep getting pushed to build for the back office,” she said. “And that’s where my heart is crying a little bit because I’m a field person.”

That tension is worth sitting with. It reveals something about how the industry’s incentive structures are wired, and who actually gets served by the current wave of construction AI.

2. 99% of Construction Data Has Zero IP in It.

The data ownership debate has been loud lately. Sarah’s take cuts through the noise. Most of what we’re fighting over. RFIs. Drawings. Submittals. Is standardized and carries almost no proprietary value. “An RFI is an RFI is an RFI,” she said. The real IP lives in scheduling logic, estimating strategy, and the interpersonal networks that keep jobsites running. It’s the superintendent who knows exactly when to call the concrete guy to make the truck show up anyway.

And the uncomfortable truth underneath all of it is that the people who hold most of that knowledge are heading into retirement, taking roughly 80% of corporate knowledge with them.

3. What AI Can Do Isn’t Necessarily What It Should Do.

This was the most important thread in the conversation. Sarah is intentionally designing Trunk Tools to keep humans in the loop, not because the technology can’t go further, but because she believes full automation of certain workflows would erode the reasoning skills the industry needs to survive.

Her submittal agent is a good example. It reads specs, compares them to incoming submittals, cross-references RFIs, and surfaces a compliance recommendation. But the final decision stays with the human. “I do not want to take the decision-making away from the humans,” she said. “But I want to take away the 95% of the bureaucratic work to get there.”

She compared it to GPS navigation. Nobody under 20 can read a paper map anymore, and that’s probably fine. But the driver still needs to know when to turn left or stop for a person in the road. The question for every AI workflow should be the same. Is this a skill we still need humans to carry? If yes, don’t automate it away.

4. The “Repetitive and Annoying” Filter Is Your Decision Framework.

Sarah has one clear rule for what to automate. If a task is repetitive and annoying, throw AI against it. If it’s not repetitive, AI probably won’t be good at it. If it’s not annoying, people actually want to do it. Leave it alone.

Nobody wants to write cost codes. Nobody wants to spend 45 minutes comparing submittals against specs. Nobody wants to fill out daily logs. These tasks are the busywork that construction has accumulated over decades. Automate those. Don’t automate reading a drawing. Don’t automate deciding when to order material. Don’t automate the decision to call the concrete guy. Those decisions are where judgment lives.

The skills lost to AI might not be skills you need to keep. You can’t hand a map to someone under 20 and expect them to read it. GPS changed that. Nobody does mental math like previous generations. Calculators ended that. We’ve been losing skills to technology for 150 years. The question isn’t whether you’ll lose skills. It’s whether the skills you’re losing matter.

5. Less Than 10% of C-Suite Executives Are Personally Using AI.

Sarah didn’t mince words here. From her conversations with construction leadership, the vast majority of executives are not hands-on with AI. They can talk about it. They can greenlight innovation budgets. But they’re not using the tools themselves. And she can tell within 10 minutes of talking to someone in an organization whether or not the C-suite is engaged.

“How do you want to change your 5,000-person organization about adopting AI if you don’t use it?” she asked.

Her point is that this isn’t just a bottoms-up training problem. It’s a top-down change management challenge. And the companies where leadership is personally experimenting. Even just analyzing their own blood work or writing stories for their grandkids. Are the ones where AI adoption filters down into real workflows.

The implication is clear. If you’re leading a company and you’re not using AI daily, your people can tell. It filters down. They’ll talk about adoption. They won’t practice it. Everything else follows from whether the person in charge is actually changed.

6. Your Competitive Advantage Is the Democratization of What Used to Be Premium Knowledge.

Here’s a shift happening right now. Large contractors used to have an advantage because they could hire McKinsey. They could afford specialized teams for everything. That advantage is evaporating.

“If I want McKinsey today, I literally open ChatGPT and tell it, please be McKinsey for me. And it does a phenomenal job,” Sarah said. Suddenly a 50-person regional contractor can access the same strategic thinking a 5,000-person national firm pays consultants for. That’s democratization.

It doesn’t level the playing field. It tilts it toward smaller, faster companies that move first. The less organizational layers you have, the faster you can make a top-down leadership change. A 100-person regional can implement AI in their workflow faster than a giant firm can change 5,000 people’s behavior.

The opportunity for the next 18 months is clear. If you’re at a regional firm, this is your moment to build capabilities that used to require big firm resources. If you’re at a big firm, this is when you lose that advantage unless you move decisively.

7. What You Should Actually Do in the Next 18 Months. By Role.

This is where it gets specific. Sarah’s advice varies depending on where you sit. The window is open now. It won’t be in 18 months.

If you’re a young person in construction (0-5 years in):

Spend 30 minutes a day. Every day. Using AI as a teacher. Not learning AI in some abstract sense. Using it to become better at your actual job.

Ask it to teach you how to be a better superintendent or PM or project coordinator. Ask it to explain why certain cost codes matter. Ask it to walk you through how to sequence a complicated phase of work. Ask it to help you think through a problem you solved on the job.

“Sit down every time when you drive home, play for 30 minutes a day with AI and let it teach you how to become a better superintendent or a better PM or a better something. Like use AI as a teacher and you will show up within a week. You can learn anything like just 30 minutes a day,” Sarah said.

Within a week, you’ll notice the difference. Not because you learned AI. But because you’re learning faster than you would learning alone. The knowledge of the whole construction site becomes available on your fingertips. That’s leverage. That’s the advantage you build before your company or your industry figures out how to systematize this stuff.

The corollary matters. She’s not telling you to train on ChatGPT or Claude. She’s telling you to use it to become better at your actual job. The tool is just the medium. The learning is the point.

And when you get stuck, remember this. Prompting is hard. Most people don’t know what they’re asking for until they start writing it down. So ask the AI to help you ask the right question. If you’re trying to understand cost codes, tell it. “I’m trying to optimize cost codes for my construction company. Can you give me a better prompt?” It sounds meta but it works. You’re using the tool to teach you how to use the tool better. That’s the real accelerant.

If you’re a superintendent or PM running projects (5-15 years in):

Your job is to identify which tasks on your projects are repetitive and annoying. Those are your AI targets. Not because AI is magic. But because it frees you to do the work that actually requires judgment.

Start small. Pick one workflow that wastes time and doesn’t develop anyone. Daily logs. Cost code entry. Submittal comparison. One task that everybody hates and nobody learns from. Bring in a tool that handles it. Deliberately keep the decision-making human. You read the AI’s recommendation. You make the call. You stay in the loop.

Do that for one project. Document what changes. What gets faster. What people learn differently. What gets worse. Then expand. But don’t automate the decision. Keep the human in the learning loop.

Also. Start using AI yourself. Not for your company yet. For your job. Use it to prepare for job meetings. Use it to think through site challenges. Use it to review your own decisions. Your crew watches. If they see you using it, they’ll adopt it. If you tell them to adopt it while you don’t use it, they won’t.

If you’re a leader at a construction company (GC, regional, mid-size):

Two things need to happen simultaneously in the next 18 months.

First. You need to start using AI daily. Not next year. Not when you hire someone to figure it out. Now. Personally. If you don’t, your people won’t. The people Sarah talks to can tell within 10 minutes if the C-suite is actually using this or just talking about it. Your people will know the same way.

Start stupidly simple. Use ChatGPT to analyze your last three monthly reports. Use Claude to help you prepare for board meetings. Use it to write an email better. Use it to think through a hire. Show your team you’re actually using this. That filters down faster than any mandate.

Second. Begin identifying your workflow bottlenecks. Where does your company waste the most time on busywork? Where are your people frustrated because they’re drowning in data instead of making decisions? That’s where you introduce AI. Not everywhere. Not yet. But start.

And make a deliberate choice about what you keep human. If your PMs are going to submit reports with AI help, decide now that they still read the output. Still make the call. Still stay in the loop. Because in five years, when you have a problem your vendor won’t help with, you need people who understand what they’re doing.

The democratization advantage means you can compete with bigger firms if you move first. But you have to actually move. Not plan to move. Not hire someone to move for you. Move.

If you’re at a large ENR firm:

Sarah was clear about this too. The “wait and see” strategy is a trap.

Many large firms want to wait for the dust to settle, to see which AI tools win before committing. Sarah argues this misses the point. Switching tools is easier than ever. The interfaces are getting simpler every month. What’s hard to recover is the two years of organizational learning your people missed while you waited.

“What’s not going to be easy is your people will miss the wave of learning slowly how to live and work with AI,” she said. The tool doesn’t matter as much as the organizational muscle you build by using one.

Your advantage isn’t that you can afford McKinsey anymore. It’s that you have deep domain knowledge and scale. But you only keep that advantage if your people are learning how to work differently right now. Not next year. Now.

Start with your innovation team. Give them real budget. Real time. Real autonomy. But also make them show results. “If a task is repetitive and annoying, you should throw AI against it,” Sarah said. Find three of those on your next major project. Automate them. Document what happens. Learn faster than your regional competition. That’s your edge. Moving faster with real insights, not bigger budgets.

8. The 18-Month Window Isn’t About the Technology. It’s About Cultural Learning.

Here’s what actually changes in 18 months. Not the tools. Not what’s possible. But whether AI adoption becomes normal or becomes a differentiator.

Right now, less than 10% of construction leaders are using AI daily. That means 90% of your competitors aren’t either. You have space to experiment. To learn. To make mistakes. To figure out what works for your specific business.

In 18 months, that flips. AI will be mainstream. Your company will mandate it. The superintendent who spent six months learning how to use it better than everyone else won’t have an advantage anymore. It’ll just be table stakes.

The people winning right now aren’t the ones who understand AI best. They’re the ones who are using it every day while their competitors are still reading articles about it. They’re building judgment about when to use it and when to think first. They’re learning what works on their specific projects instead of waiting for a playbook.

That window closes fast. After it closes, you’re playing catch-up. Not because the technology changed. But because your people didn’t learn when they had space to learn.

💬 Best Quotes

“I keep wanting to build for the field and we keep getting pushed to build for the back office. And that’s where my heart is crying a little bit because I’m a field person.”, Dr. Sarah Buchner

“An RFI is an RFI is an RFI. The real IP is in interpersonal connection between people. No AI can help with that.”, Dr. Sarah Buchner

“I do not want to take the decision-making away from the humans, but I want to take away the 95% of the bureaucratic work to get there.”, Dr. Sarah Buchner

“If a task is repetitive and annoying, you should throw AI against it.”, Dr. Sarah Buchner

“How do you want to change your 5,000-person organization about adopting AI if you don’t use it?”, Dr. Sarah Buchner

“Sit down every time when you drive home, play for 30 minutes a day with AI and let it teach you how to become a better superintendent or a better PM. You will show up within a week.”, Dr. Sarah Buchner

“If I want McKinsey today, I literally open ChatGPT and tell it, please be McKinsey for me. And it does a phenomenal job.”, Dr. Sarah Buchner

“What’s not going to be easy is your people will miss the wave of learning slowly how to live and work with AI.”, Dr. Sarah Buchner

🔚 Conclusion

This conversation with Sarah Buchner is a reminder that the most important decisions in AI aren’t about which model to use or which vendor to pick. They’re about what kind of workforce and what kind of industry we want on the other side of this transition.

The 18-month window is real. After that, this becomes normal. Your company will mandate it. The competitive advantage goes away. Right now, you can learn. You can experiment. You can build judgment about what actually works on your projects instead of following a playbook that doesn’t exist yet.

Sarah’s message is clear. Identify what’s repetitive and annoying. Automate that. Protect the reasoning. Get leadership’s hands on the tools. And stop waiting for the perfect solution while your people fall behind the learning curve.

The companies that pull ahead won’t be the ones with the best AI strategy decks. They’ll be the ones whose leaders are actually using the tools they’re asking everyone else to adopt. And whose people spent the last 18 months learning instead of waiting.